Prognostic Signature based on Transcriptome Characteristics of the C-C Motif Chemokine Receptor Genes in Hepatocellular Carcinoma and Validation

Object: This investigation aimed to assess the clinical signicance of C-C motif chemokine receptor (CCR) genes in HCC and construct the prognostic signature based on transcriptome characteristics of the CCRs. Methods: Clinical signicance of CCRs were evaluated in TCGA database and GSE14520 dataset, and prognostic CCRs (CCR1,5,7) were screened out for validation and further analysis. The relationships between CCR1,5,7 and prognosis were then evaluated in the Guangxi cohort. Based on the expression levels of CCR1,5,7 and clinicopathological characteristics, the nomograms and prognostic signatures were respectively constructed in GSE14520 dataset and Guangxi cohort. Results: CCR1,5,7 were associated with overall survival of the HCC patients in GSE14520 database, TCGA database or Guangxi cohort. In the prognostic signature, the accuracy of prognosis risk assessment based on CCR1,5,7 expression was satisfactory. The nomogram constructed in terms of the expression levels of CCR1,5,7 and clinicopathological characteristics provided a convenient tool for clinician to assess the prognostic risk of each patient. GSEA results suggested that CCRs were mainly related to B cell receptor signal pathway, chemokine signaling pathway, T cell receptor signal pathway, etc. In addition, we also found that CCR1,5,7 were signicantly positively correlated with the degree of immune inltration of B cells, T cells, and macrophages Conclusion: CCR1,5,7 might serve as prognostic biomarkers in HCC; CCR1,5,7 might regulate the progression of HCC by impacting immune cells inltration.


Introduction
There were 906,000 new cases of primary liver cancer worldwide in 2020, ranking sixth in the cancer incidence. Although liver cancer is the sixth most common malignant disease worldwide, it is the third cause of death related to malignant diseases 1,2 . Hepatocellular carcinoma (HCC) accounts for approximately 90% of all primary malignant tumors of the liver 3,4 . Cirrhosis of the liver, hepatitis B virus (HBV) and hepatitis C virus (HCV) infection, alcohol, nonalcoholic fatty liver disease (NAFLD), diabetes and obesity are the risk factors for HCC 5,6 . It has brought huge suffering, decreased quality of life and a sharp reduction in survival time to patients, and at the same time caused a tremendous economic burden to the society. The situation of HCC in China is more severe. Among the malignant diseases, the incidence of HCC ranks fourth in China and the mortality of that ranks second 6,7 . In some regions of China, such as Guangxi, due to high exposure to hepatitis B and a atoxin, the incidence and mortality of hepatocellular carcinoma has been the rst among local malignancies for four decades 7 . The Tyrosine Kinase Inhibitor (TKI) represented by Sorafenib and Lenvatinib did extend survival in some HCC patients, but the overall therapeutic effect was not satisfactory 8, 9 . Immune checkpoint inhibitors (ICI) also didn't work well in hepatocellular carcinoma alone 10,11 . The results of clinical studies in the past two years, regarding the combination of TKI and ICI, seem to offer hope for patients with advanced liver cancer 12 . The median progression-free survival (PFS) of patients taking Lenvatinib and Pabrizumab together has reached 9.7 months, and 6-month and 12-month survival rates were 83.3% and 59.8%, respectively 13 . The results of the program were considered groundbreaking. Although breakthroughs have been made in the treatment of HCC, for now, we still have a long way to go.
Chemokine receptors are known for their biological roles in chemotaxis, target cell migration, and in ammation 14 . They are not only indispensable for all protective or destructive immune and in ammatory activities, but also play an important role in the development and homeostasis of the human immune system 15,16 . Because of their important role, chemokines are closely associated with multiple diseases, such as cancer, viral infections, in ammation, and autoimmune diseases. In recent decades, chemokine system has considered as potential target for immunotherapy 17,18 . Chemokines are a large class of chemotactic cytokines, whose homologous receptors, chemokines receptors, are expressed in both tumor cells and stromal cells 19 . Given that chemokine receptors are involved in multiple aspects of cancer biology, their potential targets have been assessed in many preclinical studies and clinical trials.
Monoclonal antibodies (anti-CCR4 mAb, Mogamulizumab) and chemokine receptor inhibitors (CXCR4 antagonist AMD3100) are already being applied for hematologic malignancies in clinical 20,21 . The chemokine receptors have been grouped to subfamilies -CCR, CXCR, XCR and CX3CR -in terms of cysteine motif variations. The purpose of this investigation is to inspect the role of CCR subfamily members in HCC.

Materials And Methods
Function annotation and pathway enrichment of CCR genes Function annotation for CCRs (CCR genes) in terms of gene ontology (GO) and KEGG pathway was performed on Database for Annotation, Visualization and Integrated Discovery (DAVID) v6.8 22,23 . Function annotation clustering results were then visualized in R studio with packages GOplot 24 , Hmisc 25 , and ggplot2 26 .

Data sources and tissue specimen collection
Transcriptome sequencing data of 212 HCC patients and corresponding prognostic data in GSE14520 were obtained from GEO database 27 , with 8 patients' para-carcinoma tissues of missing. Transcriptome sequencing data of 370 HCC tissues and 50 para-carcinoma tissues were downloaded from TCGA database. In the rst a liated hospital of Guangxi Medical University, fresh liver tissues (HCC and para-carcinoma tissues) of 25 HCC patients were collected and then immersed in RNAstore Reagent (TIANGEN, Beijing). Tissue specimens were stored in the -80℃ refrigerator. All patient had signed the informed consent for investigation before operation. The study had been approved by the ethics committee of Guangxi Medical University the rst a liated hospital (Approval number: 2015 [KY-E-032]).
Expression difference analysis, correlation analysis and diagnostic e ciency Student's t test was used to analyze the expression difference of CCRs between HCC tumor tissues and para-carcinoma tissues. P<0.05 was considered statistically signi cant in Student's t test. The correlation coe cient of CCRs expression in HCC tissues was calculated in R with corrplot package. ROC (receiver operating characteristic curve) was used for assess the diagnostic e ciency of CCRs. The Area Under Curve (AUC) of the ROC curve exceed 0.70 was considered to be with satisfactory diagnostic e cacy.

Survival analysis
In GSE14520, Kaplan-Meier method and Cox proportional hazards model were respectively used for survival analysis of HCC patients in terms of expression of CCRs. Bias of clinical characteristics for survival were adjusted in Cox proportional hazards model. The CCRs associated with the OS of HCC patients in GSE14520 were integrated for combined effect survival analysis. Patients were assigned to groups based on the expression levels of multiple CCRs. Kaplan-Meier Plotter (https://kmplot.com/) is an online survival analysis website which has integrated several databases 28 . It was used to further inspect the prognostic signi cance of CCRs in TCGA database. Kaplan-Meier method was also applied for survival analysis in Guangxi cohort.

Nomogram
The nomogram was constructed in R studio with foreign package (Version 1.2.5033, R 3.6.2) in terms of clinical characteristics and expression of CCRs. Multiple value levels of each in uencing factor are assigned points, and the total score is obtained by adding the scores of each in uencing factor. The prediction probability of the individual's outcome event is calculated through the function transformation relationship between the total score and the occurrence probability of the terminal event. Bootstrap self-sampling method was used to verify the prediction e ciency of the nomogram.

Prognostic signature construction
Prognostic signature was constructed with the expression of CCRs and prognostic data. According to the regression coe cients and expression value of CCRs, risk score for each HCC patient was calculated: risk score = expression value of gene 1 x β 1 + expression value of gene 2 x β 2 +…+ expression value of gene n x β n , where β was the regression coe cient derived from the multivariate Cox proportional hazards regression model. Kaplan-Meier method was used to compare the outcome between high and low risk score groups. Time-dependent ROC curve was structured with the survivalROC package in R studio (Version 1.2.5033, R 3.6.2) to further evaluate the prediction e ciency.
Quantitative polymerase chain reaction (qPCR) Total RNA was extracted from Fresh tissues (HCC and para-carcinoma tissues) of 25 HCC patients and reversed transcribed into complementary DNA. qPCR was used to quantitatively analyze the expression of CCR1, CCR5 and CCR7 with Fast Start Universal SYBR Green Master (Roche, Germany). Primers for CCR1, CCR5 and CCR7 and GAPDH (reference gene) were synthesized by Sangon Biotech Company. The forward and reverse primer sequences of CCR1, CCR5 and CCR7 and GAPDH were as follows: Gene set enrichment analysis (GSEA) According to the median of CCR expression, the HCC patients in GSE14520/TCGA were divided into high and low expression CCR groups for GSEA. GSEA was used to explore whether there were statistical differences in Molecular Signatures Database (MSigDB) c2(c2.all.v7.0.symbols.gmt) between the genomes of high and low expression groups 29 , by virtue of standardized enrichment scores and false detection rates as criteria for determining statistical signi cance. The signi cance threshold is set to P<0.05 and false discovery rate (FDR) <0.25.

Tumor-In ltrating Immune Cells
TIMER is A Web Server for Comprehensive Analysis of Tumor-In ltrating Immune Cells 30 . It was applied for inspect the correlation between CCR genes and tumor-in ltrating immune cells in this investigation. We mainly explored the correlation between CCRs and B cells, CD8 + T cell, CD4 + T cell and macrophage. Correlation coe cient was used to evaluate the correlation between the expression level and the degree of cell invasion. The signi cance threshold is set to Correlation coe cient>0.300 and P<0.05

Statistical analysis
Student's T test was used for compare the expression difference between HCC group and para-carcinoma group. Kaplan Meier method with Log-rank test and Cox proportional hazards model was respectively applied for survival analysis. ROC analysis was performed for assessing diagnostic e ciency. Statistical calculation was implemented in SPSS 22.0 or R studio (Version 1.2.5033, R 3.6.2) except GSEA. GSEA was accomplished in software GSEA v4.0.3.
Statistical signi cance was achieved when P<0.05 in Student's t test, ROC, Log-rank test and Cox proportional hazards model. The hazard ratio was shown with a 95% con dence interval.

Function annotation and pathway enrichment result of CCR genes.
It was revealed in the gene functional enrichment analysis that CCR genes was enriched in chemotaxis, positive regulation of cytosolic calcium ion concentration, chemokine-mediated signaling pathway, immune response, dendritic cell chemotaxis, cellular defense response, and so on ( gure 1A). The correspondence between CCRs and GO terms was shown in gure 1B. The details of the enriched Gene Ontology (GO) terms in molecular function (MF), biological process (BP) and cellular component (CC) categories and KEGG pathway for CCR genes was displayed in table S1.

Expression of CCRs in HCC and para-carcinoma tissues
It was observed in GSE14520 cohort that expression of CCR1, CCR2, CCR3, CCR5, CCR7 and CCR8 in HCC tissues were signi cantly lower than para-carcinoma tissues, whereas expression of CCR6 and CCR9 was higher in HCC tissues ( gure 2A). CCR4 and CCR10 are the only two members of the CCR family that show no difference in expression between HCC and para-carcinoma tissues. Expression correlation analysis between any two members of the CCR family showed that there were strong correlations among expression of CCR1, CCR2, CCR5, and CCR7 in HCC ( gure 2B).
Then, we further evaluated the expression characteristics of CCR family genes in HCC in TCGA cohort. It was observed that expression of CCR1, CCR2, CCR4, CCR5, CCR7 and CCR9 were signi cantly lower in HCC tissues, whereas expression of CCR3, CCR8 and CCR10 were signi cantly higher in HCC tissues ( gure 2C). Expression correlation analysis indicated that there were relatively high expression correlations among CCR1, CCR2, CCR4, CCR5, CCR6, CCR7 and CCR8 in HCC ( gure 2D).

Diagnostic signi cance of CCRs in HCC
After a preliminary exploration of the expression characteristics of CCR gene family members in HCC, we assessed the possibility of these genes as diagnostic markers of HCC using the area under the ROC curve. In GSE14520 cohort, CCR1 (AUC=0.731, gure 3A) and CCR5 (AUC=0.714, gure 3E) was observed to be with good diagnostic performance in HCC, while diagnostic signi cance of the other CCR family members ( gure 3B-D, F-J) were not satisfactory. In TCGA cohort, CCR1 (AUC=0.833, gure 3K) and CCR9 (AUC=0.835, gure 3S) was found to be with good diagnostic performance in HCC, while diagnostic signi cance of the other CCR family members ( gure 3L-R, T) were not satisfactory.

Survival analysis result in GSE14520 and TCGA
In addition to whole-transcriptome microarray data and prognostic data, clinical information on 212 HCC patients was obtained from the GSE14520 dataset.
In order to adjust for the effect of clinical factors in subsequent survival analyses, we rst investigated the relationship between clinical factors and prognosis. The baseline information about the 212 HCC patients was displayed in table S2. It revealed that tumor size, cirrhosis, BCLC stage, TNM stage and AFP were associated with the OS of HCC, and tumor size, gender, TNM stage and BCLC stage were associated with the RFS of HCC.
We analyzed the relationship between CCR family members and RFS in GSE14520 and TCGA, respectively. In GSE14520 cohort, none of CCR gene was observed to be associated with RFS of patients in HCC, neither in univariate survival analysis nor after adjustment for clinical factors in Cox proportional hazards model (table 1, gure 4A-J); However, In TCGA cohort, CCR1, CCR2, CCR4, CCR5, CCR6, CCR7, CCR8 and CCR9 were observed to be associated with RFS of patients in HCC ( gure 4K, L, N-S), while prognostic signi cance was not found for CCR3, CCR10 ( gure 4M, T).
Then we evaluated the relationship between CCR family members and OS in GSE14520 and TCGA, respectively. Prognostic signi cance of CCR1 in OS (P=0.189, table 1, gure 5A) was not observed in univariate survival analysis; however, it (adjusted P=0.044, table 1) was observed to be associated with OS in Cox proportional hazards model after adjusted for clinical factors. CCR5 (P=0.022, adjusted P=0.021, table 1, gure 5E) and CCR7 (P=0.021, adjusted P=0.039, table 1, gure 5G) were both signi cantly correlated with OS in either Cox proportional hazards model or Kaplan Meier method in GSE14520 cohort. Other members of the CCR gene family were found to be associated with OS in HCC ( gure 5B-D, F, H-J).

Nomogram and prognostic signature
Based on the prognostic signi cance of CCR1, CCR5 and CCR7 found in our above study, in order to optimize our discovery and produce a better predictive prognostic model for patients with HCC, we respectively performed combined effect survival analysis, nomogram and prognostic signature in terms of the data of GSE14520. Combined analysis of CCR1 and CCR5 in HCC showed that patients in the group with low expression of CCR1 and CCR5 had the best outcome ( gure 6A). Similarly, in other combined analyses, the patient in the group , the patients in the group c and the patients in group 3 all had the longest survival in their respective comparisons ( gure 6B-D). The grouping protocols and outcomes were listed in table 2. We observed that the differences between the best and worst groups were more signi cant in the combined analysis than in the single gene survival analysis.
We established a nomogram and a prognosis signature based on the expression levels of CCR1, CCR5 and CCR7 in GSE14520. In nomogram, the length of corresponding line segment of each variable represents its contribution degree for prognosis. The parameter with the highest prognostic contribution was BCLC stage, followed by the degree of cirrhosis. The contribution of CCR1, CCR5 and CCR7 in predicting prognosis were similar ( gure 6E). We evaluated the predictive power of the histogram by the match degree between the training group and the validation group. In the nomogram of GSE14520, there was a high degree of superposition between the self-validation cohort (red line) and training group (gray line) for predicting a 1-, 3 -, or 5-year prognosis ( gure 6F-H).
The risk score formula for prognosis signature in GSE14520 was: risk score = expression value of CCR1 x -0.278 + expression value of CCR5 x -0.348 + expression value of CCR7 x -0.306. A total of 212 patients with HCC in GSE14520 were classi ed as high-risk group or low-risk group. Ranking patients by risk score from left to right ( gure 6I, K), we observed that patients in the high-risk group had a higher concentration of individuals who reach terminal event in short term ( gure 6J). The difference between the high and low risk groups in OS was statistically signi cant (P=0.025, gure 6L). Besides, the ROC curve also revealed that the prognostic signature worked well in predicting 1-, 2-, 3-, 4-, and 5-year outcome ( gure 6M).
Validation for clinic signi cance of CCR1,5,7 in Guangxi cohort Twenty-ve patients from the Department of Hepatobiliary Surgery of Guangxi Medical University were enrolled as a validation cohort. The baseline data of the HCC patients in Guangxi cohort are listed in Table S3. In Guangxi cohort, IHC assay and qPCR assay showed that CCR1, CCR5 and CCR7 expression were signi cantly decreased in HCC tissues ( gure 7A, B). Meanwhile, we observed that the expression levels of CCR1, CCR5 and CCR7 were strongly correlated ( gure 7C). Besides, it was observed in Guangxi cohort that CCR1, CCR5 and CCR7 performed well in HCC diagnosis ( gure 7D-F). In full agreement with the results in GSE14520 and TCGA database, CCR1 (P=0.045, table 3, gure 7G), CCR5 (P=0.013, table 3, gure 7H) and CCR7 (P=0.029, table 3, gure 7I) were signi cantly associated with prognosis of HCC, and their up-regulated expression predicting a good prognosis.

Nomogram and Prognostic signature construction in Guangxi cohort.
Based on the expression of CCR1, CCR5 and CCR7, we constructed the prognostic signature and the nomogram for HCC patients of Guangxi cohort. The risk score formula in Guangxi cohort was: risk score = expression value of CCR1 x -0.845 + expression value of CCR5 x -0.117 + expression value of CCR7 x -0.129.
The risk score and the time of outcome event in HCC patients of Guangxi cohort of were displayed in the scatter plots ( gure 8A, B), and the CCR1, CCR5 and CCR7 expression pro le of these patients was shown using heat map. We observed that patients in the high-risk group had a shorter survival compared to those in the low-risk group. The results of survival analysis in the high and low risk groups indicated that the difference in prognosis was statistically signi cant ( gure 8D, P=0.006). the survival ROC curve indicated that the prognostic signature worked well in predicting 1 year OS ( gure 8E).
In the nomogram constructed in Guangxi cohort, the parameter with the highest prognostic contribution was tumor size, followed by the AFP ( gure 8F). Predictive power of the nomogram was assessed using the match degree between the training group and the validation group. In the nomogram of Guangxi cohort, there was a high degree of superposition between the self-validation cohort (red line) and training group (gray line) for predicting a 1-or 2-year prognosis ( gure 8G-H).

GSEA
After intersecting the GSEA result of GSE14520 cohort with the GSEA result of TCGA cohort, it was observed that the results of these two datasets were very similar. Some of the more representative results were presented. It revealed that CCR1 ( gure 9A, B) was associated with B cell receptor signal pathway, chemokine signaling pathway, nod-like receptor signal pathway, T cell receptor signal pathway, JAK-STAT signaling pathway, etc. CCR5 ( gure 9C, D) was associated with B cell receptor signal pathway, chemokine signaling pathway, cytokine-cytokine receptor signal pathway, T cell receptor signal pathway, tolllike receptor signaling pathway, etc. CCR7 ( gure 9E, F) was associated with B cell receptor signal pathway, chemokine signaling pathway, natural killer mediated cytotoxicity, nod-like receptor signal pathway, toll-like receptor signaling pathway, etc. We observed that these CCR genes were enriched in very similar pathways in HCC data sets, which suggested that there might be synergy between them.

Tumor-In ltrating Immune Cells
We found a signi cant association between CCR1,5,7 and immune cell in ltration.

Discussion
Due to its high incidence and fatality rate, HCC brings great suffering to patients. Early screening and prognostic biomarkers for HCC are urgently needed, which may bring hope for the prevention and treatment of HCC. In recent years, the achievements of immune research have brought a breakthrough in the treatment of HCC. As chemokine receptors, CCRs play important roles in immunity and in ammation, but there are few reports on CCRs in HCC. In this investigation, we inspected the clinical signi cance of members of the CCR gene family in multiple datasets, and further explored the possible mechanisms of CCR gene in HCC via bioinformatics tools.
We rst explored genes that are differentially expressed in HCC and para-carcinoma tissue. The differentially expressed genes in TCGA LIHC dataset and GSE14520 dataset were not completely coincident, possibly due to ethnic inconsistency between the two datasets. Hepatocellular carcinoma patients in GSE14520 were all from China, represented by the yellow race, while HCC patients in the TCGA data set were mainly Caucasian. Even so, we found some common ground from the results of the two datasets. we observed that CCR1, CCR2, CCR5 and CCR7 were signi cantly lower expressed in HCC tissues in TCGA LIHC dataset and GSE14520 dataset, compared with para-carcinoma tissues.
Furthermore, survival analysis in TCGA and GSE14520 showed that CCR1, CCR2, and CCR7 were all signi cantly associated with OS of HCC patients. Integral analysis, nomogram, and prognostic model in terms of CCR1, CCR2, and CCR7 All showed good performance in prognosis evaluation of HCC. It should be noted that high expression of CCR1 in GSE14520 was associated with good outcome, whereas high expression of CCR1 in TCGA was associated with poor prognosis. We further examined the prognostic signi cance of CCR1, CCR5 and CCR7 in patients with HCC in Guangxi. We get exactly the same tendency as the GSE14520 dataset. Hepatitis B virus exposure is the main cause of HCC in China, while NAFLD is the main cause of HCC in the United States of America. We hypothesized that CCR1 might play distinct roles in HCC with different pathogeny backgrounds.
We reviewed some reports on CCR1 in multiple cancers. It prompted that higher expression of CCR1 was correlated to better prognosis of head and neck cancer, ovarian cancer and melanoma 31 . Whereas some other report showed that higher expression of CCR1 was accompanied with worse outcome of glioma, lung cancer, renal cancer, testicular cancer 32 . Zhu M et al. found that CCL14 could Induce apoptosis of hepatocellular carcinoma cells by activating CCR1 33 . This report supported our conclusion in the GSE14520 and Guangxi cohorts. It has also been found that CCL15 induces HCC cell migration and invasion through activation of CCR1, leading to a worse prognosis 34 . Besides. CCL15 also induces CCR1 /CCR3-mediated angiogenesis on vascular endothelial cells 35 ; CCL16 also could promote angiogenesis of HCC via CCR1 activation 36 . There reports con rmed our ndings in the TCGA database. CCR1 has many ligands, including CCL2, CCL3, CCL4, CCL5, CCL7, CCL8, CCL14, CCL15, CCL16, and CCL23. The levels of chemokines are different among people with different backgrounds of HCC, and then resulting in radically different consequences, via CCR1 activation.
CCR5 was usually acknowledged as the HIV speci c binding site in T cell surface. Accompanied with the rise of immunotherapy, we begun to notice the role of CCR5 in tumors. CCR5 expression has reported to be associated with the growth of multiple cancers, including breast cancer, ovarian cancer, cervical cancer, prostate cancer, colon cancer, melanoma, Hodgkin's lymphoma, and multiple myeloma 37 . XW Wang et al. found that activation of the CCL4/CCR5 axis signi cantly induced γδ T-cells in ltration in HCC, thereby improving prognosis of HCC. Leronlimab (PRO140) is a humanized IgG4 monoclonal antibody that targets chemokine receptor 5 (CCR5). In cell and animal models, it has been demonstrated to block tumor metastasis in invasive breast and prostate cancers 38 . Although the function of CCR5 in HCC is still unknown, the CCR5/CCL5 axis was attested to be associated with chronic liver in ammation caused by a variety of pathogens and be involved in the occurrence and development of HCC 39,40 . The Human Protein Atlas (https://www.proteinatlas.org/) shown that patients with upregulated CCR5 have better outcomes in a number of cancers, including thyroid, lung, colorectal, head and neck, stomach, liver, prostate, breast and cervical cancers; However, CCR5 was found to be associated with a poor prognosis in several cancers, such as Glioma, kidney cancer and Testis cancer 31,32 .The prognosis of CCR5 in multiple cancers seems to validate the function of CCR5 in this investigation.
Hypoxia and prostaglandin E2 increase the expression level of CCR7 on cancer cells, thereby affecting cell stemness and proliferation potential 15,41−45 . In colorectal cancer cells, CCL19 activates CCR7, thereby inducing miR-206 upregulation to suppress angiogenesis, with miR-206 upregulation inhibiting ERK/MAPK-HIF-1-VEGF pathway 46 . This study was the rst to inspect the prognostic signi cance of CCR7 in HCC in multiple data sets and reach consistent conclusions. CCR7 was found to be strongly associated with better outcomes in patients with hepatocellular carcinoma.
GSEA results of CCR1, CCR5, and CCR7 were very similar, and all three were found to be related to the chemotactic function of B cells and T cells. Subsequently, we investigated the correlation between CCR1, CCR5, and CCR7 and the degree of immune cell in ltration in the tumor microenvironment. The results are consistent with the GSEA discovery. We observed that CCR1, CCR5, and CCR7 were positively correlated with the degree of B cells, CD8+ T cell, CD4+ T cell and macrophage in ltration in HCC tissues.
There were several limitations in this investigation. The sample size of Guangxi cohort in this investigation was small, and a larger sample size might make the results more reliable. This study preliminarily discussed the diagnostic and prognostic value of CCR genes in HCC, however, the function of diagnostic and prognostic biomarkers in HCC still needs to be further veri ed by experiments. We found that CCR1,5,7 were related to B cells, CD8+ T cell, CD4+ T cell and macrophage in ltration in HCC tissues. However, the mechanism of leukocyte enrichment caused by them is still unclear, and animal experiments may be needed to clarify it.

Conclusion
It was discovered that CCR1,5,7 were associated with OS of patients in HCC. CCRs were closely relevant to B cell receptor signal pathway, chemokine signaling pathway, T cell receptor signal pathway, etc. In addition, we also found that CCR1,5,7 were signi cantly positively correlated with the degree of immune in ltration of B cells, CD8 + T cell, CD4 + T cell and macrophage. We suspected that CCR1,5,7 were crucial prognostic biomarkers in HCC, and CCR1,5,7 might impact HCC by induce immune cells in ltration. Authors' contributions X Zand J N conceived and designed the manuscript; Z W, J N, C L and J L made acquisition of data; T F and X L performed data analysis. RNA extraction and qPCR were done by T L and X L. X Z wrote the manuscript, and T p and X Y guided and supervised the manuscript. All authors read and approved the nal manuscript. All authors made a signi cant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave nal approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.